Fatigue Failure in Tool Steel
A tool that fails in fatigue passed the checks the drawing asked for. The hardness is on specification, the grade is the one the application called for, and the load never exceeded the design value. The crack still started, usually at a feature nobody specified, and it grew a little on every cycle until the remaining section gave way. This page covers where a fatigue crack starts on a tool, the marking and finishing practices that decide it, the contact fatigue modes that take a roll or a gear tooth out of service, and what the two industries involved finally agreed to call each one.
Why a fatigue crack starts where nobody looked
A fatigue failure begins at a place where the local stress is higher than the nominal stress on the section, and it grows under cyclic loads whose peak stays below the yield strength of the steel. The fracture that results carries two zones, a smooth progressive one where the crack advanced cycle by cycle and a rougher final one where the remaining section tore. The size of the progressive zone against the section is a direct measure of how much load the part was actually carrying, which is why the two zones are worth photographing before anything is cut.
The location of the origin is the part of the analysis a tool shop can act on, because it is nearly always a feature somebody produced. A machining groove, a stamp mark, a grinding furrow, a sharp inside corner, an inclusion, a decarburized skin, and a corrosion pit are all notches, and a notch multiplies the nominal stress by a factor set by its depth and its radius. The table below lists the sites the source works through and pairs each one with the process check that decides whether it should be there at all.
| Initiation site | Why the stress concentrates there | What to check first |
|---|---|---|
| Machining or grinding groove | A groove is a notch, and a notch multiplies the nominal stress by a factor set by its depth and its radius | The finish callout on the drawing, and whether the last operation left a circumferential mark |
| Identification stamp in a stressed area | A cold stamp displaces metal and leaves a notch around the outline of the character | Where the heat number and the part number are applied, and how deep the impression went |
| Sharp inside corner or a rapid change of section | Load lines crowd into a small radius, so the local stress exceeds the average by the notch factor | The fillet radius on the drawing against the radius measured on the part |
| Nonmetallic inclusion or stringer | A hard inclusion is incoherent with the matrix and cannot follow its strain | The cleanliness level of the steel, and where the part sat in the bar |
| Decarburized or over-carburized layer | The layer differs from the core in hardness and in residual stress, and it carries its own stress state | The stock removed before hardening and the carbon potential of the furnace atmosphere |
| Interference fit or clamped joint | The fit is a stress concentration, and when the joint breathes it adds fretting damage on top of it | Contact pressure, and whether the parts were designed to move relative to each other |
| Corrosion pit | A pit is a notch, and it holds the electrolyte that keeps the crack moving | Whether the surface was protected, and what the deposit on the fracture contains |
| Arc burn or weld repair | Local melting and rehardening leave a brittle zone in residual tension | The repair history of the tool, and the hardness of the repaired zone |
Where a fatigue crack starts on a tool, why that feature concentrates stress, and the check that ties the site to a process step. Compiled from the failure-identification sections of the source.
Two facts about the timing set the value of the discussion. In plain fatigue, crack initiation can account for about 90 percent of the total life, so the part spends almost all of its life before there is any crack to find, and an inspection interval that looks generous on paper can be most of the life. When fretting is also present at the contact, initiation falls to about 55 percent of life and the strength reduction factor rises to between 2 and 5 or more, which is why a joint that was never designed to move is such an effective way to lose a tool early. The fretting side of that pair is covered in fretting wear in tooling.
Marking the part without starting the crack
Identification marks are applied after the tool is finished, and they are not part of the design, which is exactly why they cause trouble. The source is direct about the order of preference. Raised characters are better than indented ones, because a raised mark adds material instead of displacing it. Where the mark has to be indented, a hot process is preferred over a cold one, and a marking ink is preferred over an electric etching pencil or a vibrating engraver. All three of those indenting methods are still less abusive than a cold steel stamp. Where a stamp is unavoidable, a low stress stamp with the sharp edges of the characters removed, or a dull stamp, does the least damage, particularly when the impression is light, and the stamp belongs in a known low stress area rather than next to a fillet or a section change. Characters with long straight runs have the greatest tendency to start a crack.
| Marking method | Effect on the surface | Where it is acceptable |
|---|---|---|
| Raised character, cast or forged in | Material is added instead of displaced, so no notch is introduced | Anywhere the geometry allows a raised mark |
| Marking ink | The mark is a coating and the surface is undisturbed | Any machined surface |
| Electric etching pencil | Shallow local heating with minor surface disturbance | Low stress areas and non critical surfaces |
| Vibrating mechanical engraver | A continuous groove with a fine radius at the bottom | Low stress areas only, never across a fillet |
| Hot stamping | Metal is displaced while hot and ductile, leaving a shallow notch | Low stress areas, and preferred over cold stamping |
| Cold steel stamp | Metal is displaced cold, leaving a sharp bottomed notch with residual tension around it | Known low stress areas only, applied lightly |
| Cold coining of a large mark | The most severe of the methods in both notch depth and residual stress | Avoid on any tool that sees cyclic load |
Identification marking methods in the order the source gives them, from the least to the most damaging to the surface. Source, the identification marking rules in the article Fatigue Failures.
When the environment joins in
Corrosion fatigue is the combination of a cyclic stress and an environment where neither one alone would crack the part. Its practical signature is a group of small cracks beside the main fracture, corrosion product compacted on the surface, and an origin that is rubbed and discoloured because the compression half of every cycle pressed the two faces together. In carbon steel the cracks often start at hemispherical corrosion pits, although pitting is not a requirement, and the path is usually transgranular with a small amount of branching. The number of cracks is a usable discriminator. Several corrosion fatigue cracks often grow side by side along parallel paths, while a mechanical fatigue crack may start at several points in the same region but one of them becomes dominant before the crack has gone far.
Examination order matters, and it is easy to destroy the evidence before it is read. The first step the source gives is to strip the fracture surface with cellulose acetate tape or a plastic replicating material, so that any deposit is captured for later analysis. The part is then cleaned and degreased in an ultrasonic tank of reagent grade acetone, because acetone removes the cellulose acetate residue and also dissolves magnetic particle suspension and most liquid penetrants. Solvent applied with a brush is specifically not the method. A corrosion product at the origin can also mislead at low magnification, where globular or nodular oxide is easily mistaken for intergranular facets, so no conclusion should be drawn from the appearance alone.
The number to carry into an acceptance decision is the fatigue limit, because it does not survive the environment. The fatigue limit of mild steel is eliminated in aqueous conditions depending on the electrochemical potential, and the stress range needed to cause fracture falls progressively as the time and the number of cycles increase. That is why corrosion fatigue testing measures crack growth rates in the environment rather than a total life, and why a part that is cycled in a wet or hydrogen bearing service cannot be released against a limit measured in air.
Contact fatigue where the tool rolls or slides
Contact fatigue is the failure of two curved surfaces under repeated rolling, or rolling combined with sliding, and the stress that matters is the alternating shear stress that reverses direction as the parts move. In pure rolling that shear stress peaks slightly below the surface, so a crack can start in material that looks sound from the outside, and it reaches the surface later as a pit. When sliding is added, friction moves the peak shear stress closer to the surface and raises its magnitude, which is why a gear tooth pits next to the pitch line rather than on it, where contact is pure rolling. Micropitting removes material on a finer scale again, and a case that is too thin for the load fails at the interface between case and core, a mode the source calls subcase fatigue or case crushing.
The vocabulary on this subject was a problem for a long time, because the bearing and the gear industries used different names for the same damage. ANSI/AGMA 1010-E96 was issued to settle it, and the two changes that matter most in a report are that macropitting should be used in place of spalling and micropitting in place of peeling.
| Scale | Terms the standard keeps | Terms to drop |
|---|---|---|
| Macroscale | Macropitting, pitting, initial pitting, destructive pitting, flaking, fatigue wear, and subcase fatigue or case crushing | Spalling, which has been applied to several different mechanisms and is therefore ambiguous |
| Microscale | Micropitting, microspalling, frosting, glazing, gray staining, surface distress | Peeling, which describes neither the appearance nor the mechanism of micropitting |
Contact fatigue vocabulary as aligned by ANSI/AGMA 1010-E96, which was issued so that the bearing and gear industries would describe the same damage the same way. Source, Table 1 of the article Fatigue Failures.
For a tool steel order, the useful part of that standard is the link it makes between the mode and the factor that controls it. Three of the five modes in the table below are influenced by lubrication, including every mode that starts at the surface, and the two that start below the surface are decided by the material instead. Inclusion origin contact fatigue is the primary mode in antifriction bearings, and it is the reason vacuum treated steel is used for them, because the mode is controlled by the size and density of hard oxide inclusions and the stringers they form. Subcase fatigue is a design and heat treat problem rather than a steel cleanliness one, and the fix is more case depth or more core hardness, applied carefully so that the tooth does not become through hard and brittle.
| Mode | Where the origin sits | What controls whether it happens |
|---|---|---|
| Inclusion origin, subsurface | At a nonmetallic inclusion below the contact, above the depth of maximum alternating shear stress | The size and density of oxide and other hard inclusions, and the absence of the other modes |
| Subcase fatigue, also called case crushing | Near the interface between the case and the core | A case too thin for the radius of curvature and the load, or a core hardness too low for the contact stress |
| Point surface origin | At the contact surface, where the lubricant film stops separating the asperities | Low lubricant viscosity, a film thinner than the asperities in contact, tangential force and gross sliding |
| Geometric stress concentration | At the end of line contact, where the contact stress is highest | End contact geometry, misalignment, and deflection under the applied load |
| Micropitting | At the surface, on a scale below macropitting | Low viscosity, a thin film, loss of elastohydrodynamic pressure, and slow running speeds |
The five contact fatigue modes and the factors the source lists as controlling each one. Three of the five are influenced by lubrication, including every mode whose origin is at the surface. Source, Table 2 of the article Fatigue Failures.
What this means on a tool steel order
Fatigue is the one failure mode where the drawing, the steel and the shop practice all leave a mark that can be read after the fact, and the reading is worth doing before another tool is made to the same drawing. Four items belong in the record. The finish and the fillet radius actually produced, because a notch factor is set by geometry and not by intent. The location and the method of every identification mark, because a mark in a stressed area is a designed in crack starter. The cleanliness level of the steel, where the contact stress is high enough for inclusions to matter. And the surface treatment, because a compressive layer at the surface buys initiation life and is the standard answer where the geometry cannot be changed. The residual stress that layer changes is reported the way measuring residual stress in tooling describes, and the heat treat step that decides what the surface starts with is set out in quench cracking in tool steel. The mechanisms behind the marks on the fracture surface itself are set out in fracture surface features and what they mean, the stage by stage causes of a tool or die failure are on tool and die failure analysis, and the way a fracture is classified before it is explained is on the damage mode identification chart. Tough grades for impact and cyclic service are grouped under tool steels for cracking resistance, the impact energy a grade absorbs is recorded in Charpy impact testing of tool steel, and the fracture toughness data is on the tool steel fracture toughness chart, and the way the structure behind those numbers is examined is set out in tool steel microstructure.
Before you act on a fatigue diagnosis
A reference page, it is not an Aobo Steel specification, and the figures above are the values published for the specific samples and alloys named. Fatigue life depends on the actual stress history, the geometry, the surface condition and the environment, none of which can be recovered from a single fracture photograph. A smooth zone on a fracture is evidence of progressive growth and not by itself proof of which load produced it, and a corrosion product at the origin can be mistaken for a fracture feature at low magnification. Strip and record the surface before cleaning it, and treat any life figure as an estimate tied to the inspection interval it came from.
Source: ASM Handbook, Volume 11, Failure Analysis and Prevention, ASM International, 2002.